Interpolation of all wave fields are affected by the changes to
- the default interpolation method
- the distance computation and selection of nearest points
- the handling of missing values
In particular, the different handling of missing values in the interpolation formula leads to differences close to the coasts and ice shelves. This is demonstrated here using the model bathymetry field and considering only the differences in the missing values in fields produced by MIR and EMOSLIB.EMOSLIB ...
HRES WAM 0.125 degree reduced lat-lon grid
Significant wave height The model bathymetry has been interpolated from the 0.125 degree reduced lat-lon grid target grid to regular lat-lon grids at different resolutions using EMOSLIB and MIR . and the number of missing values in the resulting fields compared (the values themselves are not considered here).
- Differences are seen as expected around the coasts
- Other parameters may also show differences close to the ice shelves which may change from day-to-day
- For wave 2d spectra small values (below approximately 10-4) are encoded with missing values. As a result, Wave 2d spectra may show additional differences over sea points well away from the coasts and ice shelves.
The plots show the location of the points that differ:
- the red dots indicate grid points where wave field produced with MIR has a missing value (indicating a land point) while while that produced with EMOSLIB has a value (indicating a sea point)
- the blue dots indicate points where the wave field produced with EMOSLIB has a missing value (indicating a land point) while that produced with MIR has a value (indicating a sea point)
- The differences for more than 99% of grid points are less than 20Pa and in the cases presented here all differences are less than 1 hPa.
- The largest differences occur in regions of high orography with the field gradient changes most rapidly.
- Larger differences are also seen in the lower resolution target grids (0.5ºx0.5º and 1.0ºx1.0º) at the polar latitudes (90ºN and 90ºS) which arise from the different handling of the interpolation at the poles.
Target | Statistics of differences |
---|
Global | Europe | ||
---|---|---|---|
0.125x0.125 | Number of points different = 114 | ||
0.25x0.25 | Number of points different = 37 | ||
0.5x0.5 | Number of points different = 17 | ||
1.0x1.0 | Number of points different = 4 |